VolunteerShiftOpen dashboard
Connecting to the agent runtime…

Your volunteer coordinator now works nights and weekends.

VolunteerShift is an autonomous agent that schedules volunteers, sends every reminder, recovers no-shows in real time and tracks hours. Coordinators spend 22+ hours a week on this. The agent does it in the background and only surfaces when there is a real decision to make.

See it working I'd like to volunteer How it works
Shifts coordinated
Messages sent
Tool calls audited
Seats confirmed
agent audit · live streaming
Built onStrands Agents SDKAmazon Bedrock AgentCoreMistral Large 3DynamoDBAmazon SESAmazon SNSS3CloudWatch

How it works

Create the shift. Everything after that is automatic.

Five specialised agents hand a shift down a pipeline. Each one has its own tools, its own system prompt and a hook that writes every tool call to an audit table before it is allowed to run.

  1. 01

    You post a shift

    Program, time, location, skills, headcount. Thirty seconds.

  2. 02

    Agents match and invite

    Best-fit volunteers get a one-tap invitation by email or SMS.

  3. 03

    No-shows get recovered

    Fifteen minutes after start, missing people trigger replacement outreach.

  4. 04

    Hours and reports write themselves

    Check-in to check-out becomes hours, reliability and a weekly report.

The agents

A small team that never drops a shift.

Built with the Strands Agents SDK as a multi-agent graph and deployed on Amazon Bedrock AgentCore. Mistral Large 3 does the reasoning; the tools do the work.

Scheduler agent

Step 1 of 5

Reads the shift, ranks the pool by skills, availability and reliability, assigns the best fit and sends invitations.

query_volunteersmatch_volunteers_to_shiftsassign_volunteers_to_shift

Communicator agent

Step 2 of 5

Runs the 3-touch sequence: invitation, 48-hour reminder, 2-hour reminder. Email or SMS, whichever each person prefers.

send_emailsend_smslog_communication

Recovery agent

Step 3 of 5

Watches check-ins as the shift starts. When someone does not show, it finds and contacts a replacement on its own.

check_shift_coveragenotify_coordinator

Tracker agent

Step 4 of 5

Logs hours from check-in to check-out and nudges reliability scores up or down so the next match is smarter.

log_hoursupdate_volunteer_profile

Reporter agent

Step 5 of 5

Turns the week into coverage, no-show and impact numbers a board or funder actually reads.

generate_report

Guardrails on every call

  • Recipients validated before any email or SMS leaves.
  • PII patterns blocked from tool inputs.
  • Every tool call logged with input and result.
  • Agents that forget to act are re-prompted until they do.
Open the audit trail

Why it matters

Built for the coordinator with 300 volunteers and no assistant.

Mid-size nonprofits run on volunteers, but the person scheduling them is usually doing it between everything else. Reminders get missed, no-shows go unnoticed until the shift is short-handed, and the hours report is a spreadsheet nobody trusts.

  • Matches on skills, availability and a reliability score that learns from every shift.
  • Volunteers reply by email, SMS or a one-tap link. No app to install.
  • Coordinators see a single ‘needs a decision’ list instead of a wall of notifications.
  • Runs on the channels you already have: Amazon SES for email, SNS for SMS.

22+

hours per week coordinators spend on scheduling admin

3

touches per volunteer: invite, 48h, 2h

15 min

after start, the Recovery agent begins replacing no-shows

100%

of agent tool calls land in the audit trail

Watch it run on live data.

The dashboard is connected to the real agent backend. Trigger the Scheduler on a shift and follow every tool call as it happens.

Open the dashboard Read the code